Text to video vs image to video: which gives more control over the final shot

Image to video gives more control over the final shot because it anchors composition, subject and framing to concrete reference media instead of relying on prompt interpretation alone. On RelayDance, image to video attaches up to 9 reference images per request in metadata.content[], cited as @image1 to @imageN in the prompt. Text to video is faster to write but leaves layout to the model. Both use the same OpenAI-compatible endpoint and pay-as-you-go billing.

How each method controls the final shot

Image to video constrains the output using visual anchors, while text to video controls it through language only. In image to video, reference media are placed in metadata.content[] and referenced in the prompt as @image1 to @imageN, letting you lock a face, product or background. A single request accepts up to 9 reference images, 3 reference videos and 3 audio tracks, with clips up to 15 seconds. Text to video passes only model, prompt and seconds, so framing, palette and subject identity are inferred. According to the official relaydance.com/docs, 「Set base_url to https://relaydance.com/v1 and keep your OpenAI SDK」, so both methods run through one integration.

Pricing and resolution comparison

Both methods share the same per-second model rates, so choosing image to video does not change the price tier. Rates are published live at relaydance.com/models. The table below lists current Seedance 2.0 options.

For image work with gpt-image-2, image output is free and only input is billed; image-to-image starts from about CNY 0.035, and 4K and 1K output cost the same.

How to submit an image to video task

You submit reference media through the same video endpoint used for text to video, then poll for the result.

1. Create an API key at relaydance.com/console and set Authorization: Bearer YOUR_API_KEY.
2. Set base_url to https://relaydance.com/v1.
3. POST /v1/video/generations with model, prompt, seconds and metadata (ratio, resolution, generate_audio, callback_url, content[]).
4. Place reference images in metadata.content[] and cite them as @image1 to @imageN in the prompt.
5. Poll GET /v1/video/generations/{task_id} until status is succeeded or failed, or set metadata.callback_url to receive the final state by webhook.

Billing safety when iterating on shots

Iterating between text to video and image to video carries no penalty for failed attempts, because billing is pay-as-you-go per generated video. According to the official relaydance.com docs, 「Failed or errored requests are never billed」, so a rejected or errored task does not add cost while you refine framing. Payments accept USDT and Stripe card. Access is direct and OpenAI-compatible: no BytePlus enterprise account or KYC is required. Because both methods use the same endpoint, protocol and per-second rates, you can test a text prompt first, then attach up to 9 reference images to tighten control, without changing your integration or pricing tier. See relaydance.com/docs for full parameters.

FAQ

Does image to video cost more than text to video? No. Both use the same per-second model rates; for example, Seedance 2.0 720p is about $0.190 / second regardless of method.

How many reference files can one request use? Up to 9 reference images, 3 reference videos and 3 audio tracks per request, with clips up to 15 seconds.

Am I charged if a generation fails? No. 「Failed or errored requests are never billed」, according to the official relaydance.com docs.